Considerations for professionals when using artificial intelligence

Artificial intelligence (AI) is trending globally. Often praised for its unprecedented opportunities for increased efficiency and fast insights, professionals should exercise caution.

In-depth evaluations should be afforded as to how AI can impact the existing legal and regulatory requirements. We examine the key considerations for professional and management matters below.

Enhancement rather than replacement

A significant risk is to use AI in place of trained professionals, whether that be solicitors, accountants, or industry experts. To accept AI-generated output in absence of thorough evaluation will always pose risks, with AI still being in its infancy. 

Whilst AI is powerful and rapidly growing, it still lacks the sophisticated understanding of context, which is key to many tasks and decision-making.

AI is not yet a replacement for expertise as, without specified aid, it is unable to incorporate nuanced distinctions, perception, and ethical considerations.

Professionals must maintain the role of making the ultimate decision and use their professional judgment, with AI supporting that role rather than replacing it. 

The pitfalls

Accuracy

 AI has the capacity to generate seemingly credible and coherent-sounding material which is fabricated. This is an AI phenomenon known as ‘hallucination’, and can occur when the training data is incomplete, biased, contains errors in the model’s architecture, or other flaws. This can be especially hazardous in professional environments; in legal work this could lead to citing non-existing cases.

In other professional fields this could lead to misrepresentation of key elements, resulting in severe consequences not only for the client but potentially to professional reputation.

Skills deficit

Despite AI being seen as a progressive technology, this could lead to a proficiency deficit among junior professionals who use AI too much to subsidise their work. 

There is the potential for these professionals to be unable to develop core professional competencies, such as analytical thinking, article and journal writing, document production and research skills. It is vital that there exists a conscious effort to balance the integrated use of AI.

Bias and fairness issues

Historical data is the basis for which AI systems are often trained, and these can consist of inherent biases. This can create discriminatory patterns in areas such as research or contract analysis.

Many professionals have ethical obligations to make sure their work promotes equality and fairness. This is something that AI cannot efficiently do without aid.

Reduced transparency

It can often be difficult to comprehend how AI has generated such conclusions. This lack of insight can be problematic within professional industries, especially when it comes to the need to be accountable and trace the logic behind decisions. This can lead to products produced by AI being undermined when put under scrutiny.

This reinforces the need for professional input and the argument for AI to assist professionals rather than perform professional functions.

Confidentiality and data security concerns 

It is crucial to give consideration to what professionals input into AI systems. When inputting client information into cloud-based or third-party tools the confidentiality concerns are vast, and businesses should ensure that any AI systems they use comply with data protection regulations and client confidentiality professional obligations.

Where the inputted data is stored should be considered because many AI platforms use servers in multiple jurisdictions, potentially subjecting users to foreign laws and surveillance management. 

How long data is stored for is a key element to understanding professional regulation. A multitude of AI models retain information to enhance the model’s performance, but subsequently, inputted data would be stored indefinitely.

Who is able to access this data, third-party employees, contractors, and automated systems often have access to data within AI systems, enabling them to process client data, potentially without the appropriate security clearance.

Dependent on where, and how long data is retained for, alongside the AI systems training data and improvement model, this can lead to fragments of inputted data being generated in its output or influencing generated data creating further regulatory data and security concerns for businesses. 

Conclusion

AI serves as a powerful agent of change across businesses and professions alike.

Its use and integration into our roles should not be shied away from because of the risks, but be approached thoughtfully, with the recognition that professional expertise and ethical reasoning, and human judgment cannot be replaced. 

While offering a multitude of genuine benefits, taking advantage of these comes with responsibility. AI should be approached thoughtfully. That being said, adoption of AI amongst professions should not be resisted due to the risks.